UPIC-EMMA 2.0: A parallel spectral Particle-In-Cell code designed for heterogeneous architectures and PIC algorithms benchmarking
POSTER
Abstract
UPIC-EMMA 2.0 is an open source (available on GitHub) fully parallelized spectral Particle-In-Cell (PIC) code. Inherited from the time-tested UPIC framework, it utilizes the message-passing interface (MPI) for parallel processing and a multi-tasking library was implemented to enable mixed multi-tasking and MPI. The full set of Maxwell equations is computed using the Fast Fourier Transform (FFT) method and a Perfectly Matched Layers (PML) technique has been developed to damp the fields at the simulation box boundaries. Numerical schemes usually used in non-spectral PIC codes can be emulated such as the Yee scheme or other higher order finite-difference time domain (FDTD) schemes. This capability makes UPIC-EMMA a unique testbed for testing new algorithms. Initially developed for studying the interaction between a highly intense laser pulse and underdense plasmas (in boosted frame, moving antenna and numerical Cherenkov instability mitigation), it has been extended to simulate the interaction between a highly intense laser pulse and an overdense plasma (relativistic binary collisions, particle antenna) and other HEDP applications. We will present the latest physics results at the poster.
*This work is sponsored by NNSA and NSF.
Presenters
-
Michael Touati
- Univ of California - Los Angeles